Tag Archives: Catastrophe Modeling

Women’s History Month
Karen Clark: A Model of Success

By Loretta Worters, Vice President, Media Relations, Triple-I

Like many people, Karen Clark’s career was influenced by circumstances and serendipity rather than advanced planning.  In graduate school she developed a love of building computer models, leading to her first job in the research department of Commercial Union Assurance. 

“One of my first assignments was to figure out if the insurer had too much coastal exposure because they had been growing along the coastline,” said Clark.  “I started to research hurricanes and how I could potentially build a model to estimate hurricane losses.” 

That research ultimately led Clark to write her seminal paper “A Formal Approach to Catastrophe Risk Assessment and Management,” published in the Casualty Actuarial Society Proceedings, in which she argued for probabilistic models rather than the subjective rules of thumb then used in underwriting. 

“Catastrophe modeling was a game-changer because it introduced a whole new way of understanding and managing risk,” Clark explained.  “We don’t just look at worse-case scenarios, but we develop a probability distribution of potential outcomes.  What are the chances of a $1 billion versus a $10 billion hurricane loss?  You need probabilities so you can evaluate how likely you are to have a solvency-impairing event and how much reinsurance you want to purchase and for pricing the product.  You also need to know what the costs and benefits are of different mitigation strategies.  That’s what was missing prior to the catastrophe models.” 

Being Taken Seriously as a Woman in the Insurance Industry

When Clark first started out, catastrophe reinsurance was primarily written out of Lloyd’s of London.  “Lloyd’s was 100% male,” she laughed.  “I gave my first presentation in the Lloyd’s Library to about 100 male underwriters.  Not only was I a woman, but I was an American woman, and I was seven months pregnant,” she said.  “Along with that, I was carting this portable computer. Many underwriters had never seen a portable computer, much less used one. 

“After my presentation, there was silence in the room, and little interest, but that didn’t dissuade me.  I was determined to find those innovators and forward thinkers and I did find a few in Lloyd’s and in the U.S., who helped me to develop AIR’s first product, CATMAP.”

Clark said it is important early on to find those forward thinkers who believe in what you’re doing and are willing to make a commitment.  She advised women not to take no for an answer and to be good communicators.  “You always have to ask for what you want.  The worse that can happen is you get a no.” 

Clark hasn’t looked back since.  As founder of the first catastrophe modeling company, Applied Insurance Research, later AIR Worldwide, she became an internationally recognized expert in the new field of catastrophe risk modeling, revolutionizing the way insurers, reinsurers and financial institutions manage their catastrophe risk. 

Clark declined many offers to sell her company over the years, but eventually decided to sell AIR to Insurance Services Office (ISO).  Several years later, she co-founded  Karen Clark & Company (KCC) with her business partner, Vivek Basrur, never intending to develop catastrophe models again.  “But as my partner likes to say, life is what happens when you have other plans.”

Reinventing an industry

“Through numerous consulting engagements with global (re)insurers we discovered the models were not meeting all the needs of the senior level decisions makers.  We started hearing several consistent themes and eventually developed what we called the CEO Wish List”, said Clark.

That CEO Wish List informed the KCC vision for a new generation of catastrophe models—models that are more accurate, fully transparent, and provide decision makers with additional risk metrics and insight into large loss potential.  “We didn’t change the fundamental structure of the models”, says Clark, “but rather how the models are  delivered to (re)insurers and how they can be leveraged in new ways.”

Clark said that KCC is doing a few things differently than other modelers and one of them is their scientific approach.  “Rather than extrapolating from historical data, we have implemented advanced physical modeling techniques for the more frequent events, such as severe convective storms, winter storms, and extratropical cyclones.  This enables our models to capture all weather-related claims and not just those defined as catastrophes.  Our internal systems automatically ingest over 30 gigabytes of data a day from all the satellites, radar stations and global models so our clients have high resolution hazard footprints every morning for monitoring and managing daily claims activity. 

“Interestingly, reinventing the catastrophe modeling industry was just as challenging as inventing it”, says Clark, “because most people thought it was impossible.”  “We again had to find those industry leaders and early adopters who believed in our vision and then worked with us to make it a reality.”

Clark said she’s very fortunate she discovered her passion at a young age when she first started her career.  I just love what I do, and until I can come up with something else that I could enjoy doing daily as much as I enjoy KCC, I’ll be right here.”

Cross-posted from the Triple-I Resilience Accelerator blog

New insurance advisory board seeks technological solutions to disaster resilience

The insurance industry continues to be a major stakeholder in mitigating the effects of natural disasters on communities. As such, a group of U.S. insurers, reinsurers, intermediaries, and model providers are creating an advisory board called Helix.

Facilitated by The Institutes, Helix seeks to integrate new approaches to automated claims analysis into an overarching framework for the application of new and emerging technologies in natural disaster resilience, according to a Risk & Insurance article.

“We are excited to help coordinate this effort focused on mitigating the adverse effects of natural disasters,” says Peter Miller, President and CEO of The Institutes. He described Helix as an opportunity “to serve as a neutral third party in work on this important issue that ultimately benefits the general public.”

Initially building on work to implement open common data standards for catastrophe risk analytics, the Helix vision is grounded on four pillars to support the industry’s increasingly wide-ranging and growing capabilities:

  • Climate and resilience: Pursuing hazard and resilience research and advocating for innovation in insurance products and economic responsiveness;
  • Data standards, data content/interpretation/quality, and industry-level data resources;
  • Technology: Transparency in models and analytics, Insurtech innovations, and technology solutions;
  • Operations: Common industry tools, improved communication/exchange across the value chain, and support/education for the industry

Helix builds on the work of The Institutes’ Catastrophe Modeling Operating Standards (CMOS) initiative. The CMOS team completed a survey project in September 2020 to establish and implement an open common exposure data standard. This project also provided a set of recommendations for the community to advance the work.

“Based on the interest in and success of the CMOS, it is clear there is a desire for an industry-wide, cooperative effort focused on resilience from natural catastrophes,” says Sean Ringsted, Chief Risk Officer, Chubb. “We’ve received strong interest in creation of Helix and look forward to welcoming the participation of additional organizations.”

The Institutes is in the process of engaging founding members and building out the appropriate governance structure. As those are put in place, Helix members will determine initial priorities in support of the four pillars and leveraging the work performed under the CMOS initiative. Companies in search of additional information, or that have interest in contributing expertise to the effort can contact The Institutes at helix@theinstitutes.org.

Cross-posted from the Triple-I Resilience Accelerator

Private market flood insurance is cheaper in many cases

Alongside the National Flood Insurance Program (NFIP), a thriving private flood insurance market would provide wider and in many cases cheaper coverage options, according to a new study.

Consulting firm Milliman, in partnership with risk modeler KatRisk, looked at three states – Florida, Texas, and Louisiana – which combined account for 56 percent of NFIP insurance policies in-force nationwide.

Its analysis compared modeled private flood insurance premiums to those of the NFIP.

Key findings:

  • Some 77 percent of single-family homes in Florida, 69 percent in Louisiana, and 92 percent in Texas could see cheaper premiums with private insurance than with the NFIP.
  • Of the homes modeled, 44 percent in Florida, 42 percent in Louisiana and 70 percent in Texas, could see premiums that are less than one-fifth that of the NFIP.
  • Conversely, private insurance would cost over twice the NFIP premiums for 14 percent of single-family homes in Florida, 21 percent in Louisiana and 5 percent in Texas.

prior post discussed how private carriers are dipping their toes in the flood insurance market.

Private Market Flood Insurance Is Budding

Private carriers are dipping their toes in the turbulent waters of flood insurance, writes Insurance Information Institute (I.I.I.) research manager Maria Sassian.

This year, for the first time, insurers were required to report in their annual statements data on private flood insurance.

I.I.I. has compiled a list of top insurers in the market by 2016 direct premiums written, based on data from S&P Global Market Intelligence:

As you can see, the top three companies hold almost 81 percent of the market share, and at number one FM Global has a 54 percent market share. Direct premiums written for all companies total $376 million.

Private flood includes both commercial and private residential coverage, primarily first-dollar standalone policies that cover the flood peril and excess flood. It excludes sewer/water backup and the crop flood peril.

Some of the reasons private insurers are becoming more comfortable covering flood risk include: improved flood mapping technology; improved flood modeling; the construction of flood resistant buildings; and encouragement from Congress.

The Federal Emergency Management Agency’s National Flood Insurance Program (NFIP) is billions of dollars in debt due to large losses from Hurricanes Katrina, Rita and Superstorm Sandy. Opening the market to private insurers is one of several measures enacted by lawmakers to get the program out of debt.

Another step in shoring up the NFIP took place with the January 2017 transfer of over $1 billion in financial risk to private reinsurers. FEMA gained the authority to secure reinsurance from the private reinsurance and capital markets through the Biggert-Waters Flood Insurance Reform Act of 2012 and the Homeowners Flood Insurance Affordability Act of 2014 (HFIAA).

What Does Private Market Flood Insurance Look Like?

In his second post from the Cat Risk Management 2017 conference, Insurance Information Institute chief actuary James Lynch discusses private market flood insurance options:

Florida has opened its market to private flood insurance, and there has been some activity in that area. Most plans have been National Flood Insurance Program (NFIP) clones in that they mimic how the NFIP prices risk but introduce a lot of underwriting rules to try to avoid problem risks.

Other than mimicking the NFIP program, there are two alternative ways to price risk:

    • Develop a refined rating plan, which resembles (to me at least) a traditional classification plan. The company develops a base rate then credits and debits a risk based on factors like:
      • Elevation.
      • Relative elevation (whether a risk is higher or lower than the areas that immediately surround it).
      • Distance to coast.
      • Distance to river.
    • Use a sophisticated catastrophe model to price each risk individually. That approach is more precise, but it could be more difficult to pass regulatory approval.  (The model might be too much of a black box.) It could also be harder for agents to understand the model and explain it to clients.

Much of the industry long-term seems interested in how computer models can price flood risk, but most people recognize the challenges. A big one is how to build in the precision necessary.

Figuring out how far a property is from a river is easy. But it is hard to use Big Data techniques to determine something as simple as whether a property has a basement; let alone knowing the elevation of the lowest vulnerable point in a property. (Hint: It’s probably not the front threshold.)

Private Market Looks Closely At Flood Insurance

Almost all private insurers have shunned covering flood since the 1950s, but that could be changing fast, writes Insurance Information Institute (I.I.I.) chief actuary James Lynch:

At the Cat Risk Management 2017 conference I attended earlier this month, flood was the hottest topic. Here’s why:

  • Insurers have become increasingly comfortable with using sophisticated models to underwrite insurance risk, and modeling firms are getting better at predicting flood risk.
  • The federal government, which insures the vast majority of flood risk, is looking for ways to share the risk with private industry. Key reasons:
    • The National Flood Insurance Program (NFIP) owes the Treasury more than $20 billion (thanks to flooding from Hurricane Katrina and superstorm Sandy). It has no practical way to pay that back, and the government has made it clear that it doesn’t want to fund more losses. So the NFIP is purchasing private reinsurance. More on that below.
    • The number of people who lack flood insurance is distressingly high. I.I.I. surveys show that only about 12 percent of Americans have flood insurance. The government wants people to be protected, and encouraging a private flood insurance market could do that.

Here are some of my notes from #catrisk17 on flood insurance:

  • The NFIP reinsurance deal (effective January 1, 2017) means that reinsurance would reimburse NFIP for 26 percent of the losses from an event where losses exceed $4 billion. The maximum recovery is $1.046 billion, and the cost, according to my notes, is $150 million. (If you work in reinsurance it may be easier to think of the pricing this way: NFIP cedes 26 percent of the $4 billion excess $4 billion occurrence layer at a 14.3 percent rate on line.) There have only been a couple of floods that big in NFIP history (Hurricane Katrina and superstorm Sandy), so the cover is in place primarily to protect against storm surge. However, it would cover other major types of flood as well.
  • A significant obstacle to modeling flood risk is the fact that much of the most important data (underwriting and claims information) is in the federal government’s hands. The government wants to share the data responsibly, but its hands are tied by federal rules on sharing data about individuals. The rules are driven both by privacy concerns and cyber security laws. The government will likely be developing a certification process so that professionals could qualify to have access to the data on a limited basis.
  • A live poll found that flood modeling was the most important topic at the conference, cited by 56 percent of respondents – outpacing severe convective (thunder) storm models, cyber insurance models or terrorism models.

From Many Models, One Decision

Insurance Information Institute chief actuary James Lynch previews one of the most important conferences in the catastrophe modeling world.

I will be attending Cat Risk Management 2017 in Orlando next week, and the reason is as close as the weather forecast I’m looking at early Wednesday.

By now, the weather models have more or less converged: my own sliver of New Jersey is forecast to get about 6 inches of snow. The key word in that last sentence is models.

The many organizations that forecast the weather – the Weather Channel, Accuweather, Weather Underground, the National Weather Service – even the hearty jokester on your local station – use multiple models to predict sun, rain or snow.

The similarity to actuarial work is striking. Like an actuary, the weatherman hasn’t built the models but has to understand the strengths and weaknesses of each. And she has to make a single, certain prediction, yet couch that certainty within a pocket of doubt. The National Weather Service predicts 6.7 inches for my hometown: as much as 7 but as little as 3 (Editor’s Note: total snowfall 6.3 inches by Thursday evening).

Actuaries do that with your insurance policy – many uncertainties but one price. Of the many risks with which they must contend is how their portfolio of policies will perform under a catastrophe. Years ago this risk was estimated crudely – the old Casualty Actuarial Society exams included a section on the ISO Excess Wind calculation. Now catastrophe models do the job. And insurers need a lot of catastrophe models, which is what will be taking me to Orlando.

Next week’s conference is a cornucopia of cat models – hurricane models and wildfire models, earthquake and flood models. There is even discussion of how to coordinate the many models insurers must juggle. The conference, presented by the Reinsurance Association of America, is sold out; about 500 will attend.

I will be live-tweeting and will post a report. I.I.I. wants to draw attention to the importance of resilience – helping people understand that the best way to rebound from cataclysm is to prepare for it. Explaining how insurers do their part – in this case using models so that a policy’s price reflects its risk – helps everyone understand how much risk they must prepare for.

And I suppose, yes, will be good to visit balmy Florida after digging out from a half-foot of snow.

I.I.I.’s Facts and Statistics on global catastrophes gives a good idea of the scope of disasters that insurers protect against.